ECSIT Is a Critical Limiting Factor for Cardiac Function

ECSIT is a critical limiting factor for cardiac function

Linan Xu, … , David J. Grieve, Paul N. Moynagh

JCI Insight. 2021. https://doi.org/10.1172/jci.insight.142801.

Research In-Press Preview Cardiology Metabolism

Graphical abstract

Find the latest version: https://jci.me/142801/pdf 1

2 ECSIT is a critical limiting factor for cardiac function

1 1 1 1 4 Linan Xu , Fiachra Humphries , Nezira Delagic , Bingwei Wang , 1 2 2 5 Ashling Holland , Kevin S. Edgar , Jose R. Hombrebueno , Donna 3 1 1 6 Beer Stolz , Ewa Oleszycka , Aoife M. Rodgers , Nadezhda 4 5 2 7 Glezeva , Kenneth McDonald , Chris J. Watson , Mark T. 5 2 2 8 Ledwidge , Rebecca J. Ingram , David J. Grieve and Paul N. 1,2,6* 9 Moynagh

10 1The Kathleen Lonsdale Institute for Human Health Research, Department of

11 Biology, Maynooth University, Maynooth, Co. Kildare, Ireland. 2Wellcome-

12 Wolfson Institute for Experimental Medicine, Queen's University Belfast,

13 Belfast, United Kingdom. 3Center for Biologic Imaging, University of Pittsburgh

14 Medical School, Pittsburgh, PA 15261, USA. 4Conway Institute, University

15 College Dublin, Belfield, Dublin, Ireland. 5Chronic Cardiovascular Disease

16 Management Unit and Heart Failure Unit, St Vincent's Healthcare Group/St

17 Michael's Hospital, County Dublin, Ireland

19 6Lead Contact: P.N.M.

20 *Correspondence:

21 Paul Moynagh (E: [email protected]; Tel: 00353-1-7086105)The Kathleen Lonsdale 22 Institute for Human Health Research, Department of Biology, Maynooth University, 23 Maynooth, Co. Kildare, Ireland.

24 The authors have declared that no conflict of interest exists

26 Abstract 27 ECSIT is a protein with roles in early development, activation of the transcription factor 28 NFB and production of mitochondrial reactive oxygen species (mROS) that facilitates 29 clearance of intracellular bacteria like Salmonella. ECSIT is also an important assembly factor 30 for mitochondrial complex I. Unlike the murine form of Ecsit (mEcsit), we demonstrate here 31 that human ECSIT (hECSIT) to be highly labile. In order to explore if the instability of hECSIT 32 affects functions previously ascribed to its murine counterpart, we created a novel transgenic 33 mouse in which the murine Ecsit gene is replaced by the human ECSIT gene. The humanised 34 mouse has low levels of hECSIT protein in keeping with its intrinsic instability. Whereas low 35 level expression of hECSIT was capable of fully compensating for mEcsit in its roles in early 36 development and activation of the NFB pathway, macrophages from humanised mice showed 37 impaired clearance of Salmonella that was associated with reduced production of mROS. 38 Notably, severe cardiac hypertrophy manifested in ageing humanised mice leading to 39 premature death. The cellular and molecular basis to this phenotype is delineated by showing 40 that low levels of human ECSIT protein leads to marked reduction in assembly and activity of 41 mitochondrial complex I with impaired oxidative phosphorylation and reduced production of 42 ATP. Cardiac tissue from humanised hECSIT mice also shows reduced mitochondrial fusion 43 and more fission but impaired clearance of fragmented mitochondria. A cardiomyocyte- 44 intrinsic role for Ecsit in mitochondrial function and cardioprotection is also demonstrated. We 45 also show that cardiac fibrosis and damage in humans correlates with low expression of human 46 ECSIT. In summary, our findings identify a new role for ECSIT in cardioprotection whilst also 47 generating a valuable new experimental model to study mitochondrial dysfunction and cardiac 48 pathophysiology.

49 Keywords: Cardiac Hypertrophy/ECSIT/Mitochondrial Complex I

51 Introduction 52 Toll-like receptors (TLRs) recognise pathogen-associated molecular patterns (PAMPs) 53 and respond by inducing the expression of pro-inflammatory proteins (1). The engagement of 54 TLRs by PAMPs leads to formation of receptor proximal signalling complexes consisting of 55 TIR domain-adaptor proteins, IRAK kinases and the E3 ubiquitin ligase TRAF6 (2). Activation 56 of the latter results in its auto-ubiquitination and formation of unanchored polyubiquitin chains, 57 leading to recruitment of Tak1 kinase and downstream activation of the transcription factor 58 NFB and MAP kinase pathways that drive inflammatory gene expression (3). Evolutionary 59 Conserved Signalling Intermediate in Toll pathway (Ecsit) was first described as a positive 60 regulator of NFκB by interacting with TRAF6 (4) and more recent reports also indicate its 61 interaction with Tak1 kinase (5) and NFB proteins (6). A mutant form of Ecsit, that strongly 62 activates NFκB, has been shown to drive hyperinflammatory disease (7). Other studies have 63 identified and characterised Ecsit as part of Complex I in the electron transport chain of 64 mitochondria (8-11). An N-terminal mitochondrial localisation sequence directs Ecsit to the 65 mitochondria to facilitate assembly of complex I. Furthermore in macrophages infected with 66 intracellular bacteria such as Salmonella Typhimurium (S. Typhimurium) Ecsit can interact 67 with TRAF6 to facilitate the juxtaposition of mitochondria and phagosomes and augment 68 reactive oxygen species (ROS) production to enhance bactericidal activity (12, 13). In 69 macrophages, Ecsit also mediates recruitment of phagosomes to damaged mitochondria to 70 promote removal of the latter by mitophagy (14).

71 Most analysis of Ecsit function has focused on the murine form of ECSIT (mEcsit) and at the 72 cellular level. Interestingly, across all human tissues, human ECSIT (hECSIT) is expressed at 73 its highest in heart but efforts to study the broader physiological role(s) of Ecsit, beyond innate 74 immunity, have been hampered by deletion of the Ecsit gene in mice resulting in early 75 embryonic lethality due to impaired BMP signalling during mesoderm formation (15). As part 76 of our efforts to study hECSIT, we noted early in our studies that the hECSIT protein is more 77 labile than its murine counterpart. To understand the physiological relevance of this differential 78 lability, humanised mice were generated in which the mEcsit gene was replaced by the hECSIT 79 gene. Tissues from humanised hECSIT mice displayed reduced levels of Ecsit but mice 80 survived and developed normally to adulthood. Macrophages from these humanised mice 81 showed normal activation of the NFB and MAP kinase pathways but reduced production of 82 mROS and impaired bacterial clearance in response to challenge with S. Typhimurium. Signs 83 of cardiac hypertrophy were apparent at a young age in the humanised hECSIT mice and

84 progressed to severe hypertrophy and heart failure in later life. This pathology was associated 85 with marked reduction in subunits and assembly of mitochondrial complex I with impaired 86 oxidative phosphorylation and reduced production of ATP. Cardiac tissue from humanised 87 hECSIT mice also showed reduced mitochondrial fusion and more fission but impaired 88 clearance of fragmented mitochondria. The effects of Ecsit depletion on cardiac hypertrophy 89 was also reproduced at a cell level in human cardiomyocytes confirming a cardiomyocyte- 90 intrinsic role for Ecsit in mitochondrial function and cardioprotection. In human cohorts, we 91 also demonstrate an inverse correlation between levels of hECSIT in heart tissue and cardiac 92 fibrosis and disease. This study adds further understanding to the recognised importance of 93 mitochondrial dysfunction being a major contributor to impaired cardiac functions (16). 94 Pathogenic mutations in core protein subunits and assembly factors of complex I, leading to 95 impaired complex I function, have been reported in a number of diseases including heart failure 96 and cardiomyopathy (17-19). This study now identifies hECSIT as a critical limiting factor for 97 mitochondria function and cardioprotection.

100 Results

101 Human ECSIT is more labile than murine Ecsit

102 As part of our efforts to study hECSIT, we initially noted intriguing differences in levels of 103 murine and human forms of ECSIT when we overexpressed myc-tagged forms of both proteins 104 in HEK293T cells (Supplemental Figure 1A, upper panels). The steady state expression levels 105 of hECSIT protein were consistently less than its murine counterpart. Whereas overexpression 106 of mEcsit was sufficient to induce expression of a co-transfected NFB-regulated reporter 107 gene, hECSIT was ineffective (Supplemental Figure 1A, lower panels). The inability of 108 hECSIT to activate NFB may be due to a potential functional difference from mEcsit or 109 alternatively its low level of expression. Indeed, mEcsit was more resistant to degradation than 110 hECSIT when cells were treated with the protein translation inhibitor cycloheximide 111 suggesting greater stability for mEcsit (Supplemental Figure 1B). Both proteins migrate as 2 112 immunoreactive bands with the slower migrating form of hECSIT being especially susceptible 113 to degradation. Both proteins were expressed at comparable levels in extracts from an 114 inducible prokaryotic expression system (Supplemental Figure 1C) but hECSIT was less stable 115 during purification to homogeneity with greater amounts of purified mEcsit being isolated from 116 similar scales of bacterial cultures (0.75 mg hECSIT versus 1.9 mg mEcsit from 1L culture, 117 Supplemental Figure 1D). We also compared the relative stabilities of both purified proteins 118 by characterising resistance to proteolytic processing by the non-specific protease Proteinase 119 K (Supplemental Figure 1E). Whilst both proteins were sensitive to processing, mEcsit was 120 more resistant suggesting that the folding structure of mEcsit make it more refractory to 121 degradation by proteolysis. The anti-ECSIT antibody showed the same level of 122 immunoreactivity with equal amounts of purified mEcsit and hECSIT (Supplemental Figure 123 1F) confirming that western blotting analysis with this antibody can differentiate between 124 varying levels of mEcsit and hECSIT and excludes the possibility of differential 125 immunoreactivity of the 2 forms of protein with the antibody.

126 Given the more labile nature of hECSIT coupled to its inability to activate NFB we were keen 127 to explore the physiological relevance, if any, of these differences with mEcsit. We thus created 128 a humanised mouse in which the mouse Ecsit gene was replaced by the human ECSIT gene 129 under the control of the endogenous murine Ecsit promoter. Homologous recombination was 130 used in embryonic stem cells to replace the murine coding exons 3-9 with the human coding 131 exons 2-8 (Supplemental Figure 2A), as confirmed by PCR analysis (Supplemental Figure 2B).

132 Heterozygous knock-in mice (ECSIT+/-) containing a copy each of the murine Ecsit and human 133 ECSIT alleles were viable and these mice were inbred to generate homozygous knock-in mice 134 (ECSIT+/+) containing 2 copies of the human ECSIT allele. The crosses of heterozygous mice 135 resulted in wild type, heterozygous and homozygous mice with frequencies predicted by 136 Mendelian genetics (Supplemental Figure 1C). The homozygous human ECSIT knock-in mice 137 are viable and develop normally in early age. All studies used littermate mice derived from 138 crossing of heterozygous human ECSIT knock-in mice. Since previous studies have shown that 139 knockout of the mouse Ecsit gene results in early developmental lethality due to loss of BMP 140 signalling (15) the survival and development of homozygous human knock-in mice containing 141 the human ECSIT alleles clearly indicates that the human ECSIT protein can functionally 142 substitute for the murine equivalent in this early developmental pathway. We next examined if 143 hECSIT could similarly fulfil the other previously described roles for mEcsit in activation of 144 the NFB pathway and clearance of intracellular bacteria.

145 Bone marrow derived macrophages from hECSIT knock-in mice show normal activation 146 of NFB and pro-inflammatory cytokine expression but impaired mROS production and 147 bacterial clearance

148 BMDMs were prepared from ECSIT+/+ mice and again cellular protein levels of hECSIT were 149 lower than mEcsit from WT mice (Figure 1A). However, ECSIT+/+ BMDMs showed similar 150 responsiveness as WT cells to LPS in terms of temporal activation of the NFB and MAPK 151 pathways (Figure 1A) and induction of the pro-inflammatory cytokines IL-6 (Figure 1B) and 152 TNF (Figure 1C). Furthermore, infection of ECSIT+/+ and WT BMDMs with Salmonella 153 Typhimurium showed comparable activation of NFB and MAPK pathways (Figure 1D) and 154 induction of IL-6 (Figure 1E) and TNF (Figure 1F). These findings suggest that lower levels 155 of hECSIT do not limit the early signalling pathways and pro-inflammatory cytokine 156 expression triggered by LPS and TLR4. The inability of overexpressed hECSIT to activate 157 NFB in 293T cells (Supplemental Figure 1A) suggests a cell type specific role for hECSIT in 158 this pathway or alternatively ECSIT is required but not sufficient for activation of NFB and 159 may require other cellular components to be triggered by LPS and the TLR4 pathway. However 160 BMDMs from ECSIT+/+ mice displayed a tendency towards reduced mROS but not cROS, 161 relative to WT cells, in response to infection by S. Typhimurium (Figure 1G). This suggests 162 that the lower levels of hECSIT are critical to the mitochondrial function of Ecsit at least in the 163 context of mROS production. Given the importance of Ecsit in promoting mROS production

164 to effect bactericidal activity against S. Typhimurium we also investigated the capacity of cells 165 from ECSIT+/+ mice to clear S. Typhimurium. At higher multiplicity of infection (MOI), 166 BMDMs from ECSIT+/+ mice displayed increased bacterial burden, relative to WT BMDMs, at 167 24 h post-infection by S. Typhimurium (Figure 1H). The physiological relevance of this 168 limiting aspect was further confirmed in an in vivo infection model in which ECSIT+/+ mice, 169 orally infected with Salmonella, demonstrated impaired clearance and increased dissemination 170 to the liver and spleen when compared with WT and ECSIT+/- mice. (Figure 1I).

171 Human ECSIT knock-in mice develop early cardiac hypertrophy leading to premature 172 death.

173 Whilst ECSIT+/+ mice showed normal development and maturation into adulthood, we noted 174 that as ECSIT+/+ mice aged, they were subject to premature death from heart failure with some 175 mice dying at 8 months of age and less than 20% of the mice being viable at 12 months (Figure 176 2A). The viability of ECSIT+/- mice over this period was comparable with WT mice. Post- 177 mortem analysis displayed very distinct enlargement of hearts in ECSIT+/+ mice relative to age- 178 and sex-matched WT mice (Figure 2B). The greater heart:body weight ratio was evident from 179 8-10 weeks of age and further increased at 7 months (Figure 2C and Supplemental Figure 180 S3A). Tissue enlargement in ECSIT+/+ mice was restricted to the heart since other tissues such 181 as lung, liver, spleen and kidney showed the same size and weight in WT and ECSIT+/+ mice 182 (Figure 2C and supplemental Figure 3A). Notably, heart enlargement was only apparent in 183 homozygous ECSIT+/+ and not heterozygous ECSIT+/- mice (Figure 2C and supplemental 3A). 184 Echocardiographic analysis was performed on 6-7-month-old mice in order to further explore 185 and define the effects of knock-in of human ECSIT on cardiac physiology. Echocardiography 186 confirmed marked cardiac hypertrophy (increased LVPWS, LVPWD; Supplemental Table 2) 187 together with significant chamber dilatation (increased LVEdD; Figure 2D) and systolic 188 (increased LVEsD, decreased FS; Figure 2D) and diastolic dysfunction (increased IVRT and 189 MPI; Supplemental Table 2) in ECSIT+/+ but not WT or heterozygous hearts, despite 190 compensatory increases in heart rate observed in ECSIT+/+ mice (Figure 2D). Notably, these 191 major structural and functional alterations were associated with significant ECG changes, 192 specifically prolonged P-R interval and QRS complex duration (Supplemental Table 2), 193 suggestive of electrophysiological dysfunction. Histological analysis of cardiac tissue using 194 H&E staining highlighted abnormal tissue architecture in humanised hearts (Figure 2E). Wheat 195 germ agglutinin (WGA) staining of cell membrane proteins (Figure 2E) allowed for calculation

196 of cardiomyocyte cross-sectional area (Figure 2F) and confirmed cardiac hypertrophy at the 197 cellular level, in cardiac tissue from ECSIT+/+ mice . Heart tissue from ECSIT+/+ also displayed 198 clear signs of damage and fibrosis as indicated by greater Trichrome staining of collagen 199 volume (Figure 2E-2F). ECSIT+/+ heart tissue also showed more intense immunostaining of 200 extracellular matrix proteins laminin and collagen-VI and the fibrotic marker periostin 201 (Supplemental Figure 3B). In an effort to understand if cell apoptosis underpinned the cell 202 damage, we performed TUNEL (terminal deoxynucleotidyl transferase-mediated deoxyuridine 203 triphosphate nich end labeling) staining (Supplemental Figure 4A-4B) and analysed levels of 204 cleaved caspase 3 (Supplemental Figure 4C) However TUNEL staining and caspase 3 cleavage 205 were comparable in heart tissue from WT and ECSIT+/+ mice suggesting that necrosis rather 206 than apoptosis is the likely mechanism of any cardiomyocyte cell death in ECSIT+/+ mice.

207 Given our earlier findings of reduced stability of hECSIT relative to mEcsit, we assessed the 208 expression levels of the murine and human forms of these proteins in heart tissue from WT and 209 ECSIT+/+ mice respectively. Western blot analysis clearly demonstrated greatly reduced levels 210 of hECSIT in ECSIT+/+ heart relative to levels of mEcsit in WT heart with ECSIT+/- heart 211 displaying expression of both forms (Supplemental Figure 3C, lower panel). Interestingly 212 quantitative RT-PCR analysis demonstrated comparable levels of mRNAs encoding mEcsit 213 and hECSIT in cardiac tissue from WT, ECSIT+/- and ECSIT+/+ mice consistent with our 214 proposed model of hECSIT protein being less stable than mEcsit (Supplemental Figure 3C, 215 upper panel). The decreased levels of hECSIT relative to mEcsit was apparent in hearts from 216 male and female mice (Supplemental Figure 3D). The relatively lower levels of hECSIT 217 protein relative to mEcsit was also apparent in other tissues from male and female mice 218 including spleen, lung, liver and kidney although the biggest difference in levels was observed 219 in the heart (Supplemental Figure 3E). All of these data are consistent with hECSIT being less 220 stable than mEcsit with the reduced steady state levels of hECSIT underlying cardiac 221 hypertrophy and fibrosis and threshold levels being required for normal cardiac function.

222

223 Low levels of human ECSIT correlates with cardiac hypertrophy and fibrosis

224 Given that low levels of hECSIT manifest in cardiac hypertrophy, we are keen to examine if 225 this relationship had pathophysiological relevance in man. Intriguingly gene expression 226 analysis across 86 human tissues/cell types showed hECSIT to be expressed at its highest level 227 in the heart (Supplemental Figure 5; http://biogps.org/#goto=genereport&id=51295) (20-23).

228 We next examined ex-vivo human cardiac tissue samples from patients undergoing cardiac 229 surgery and explored the association of hECSIT gene and protein expression in the context of 230 human cardiac hypertrophy and fibrosis. Briefly, cardiac tissue samples were obtained from 231 the right atrial appendage adjacent to the venous cannulation site in 38 consecutive, consenting 232 patients undergoing elective surgery for coronary artery bypass grafting or valve replacement. 233 In addition, serum samples for biomarker analyses and Doppler Echocardiographic images 234 were obtained. Patients were aged 68.0 ± 9.6 years, 28 (74%) were male, 27 (71%) had 235 ischaemic heart disease, 19 (47%) had valvular heart disease and 16 (42%) had left ventricular 236 hypertrophy (>95 g/m2 for women and >115 g/m2 for men (24). Patients did not have evidence 237 of overt heart failure and most patients had a normal ejection fraction (EF > 50%). Other patient 238 characteristics, clinical demographic and Doppler Echocardiographic criteria are presented in 239 Supplemental Table 1, according to the presence or absence of evidence of left ventricular 240 hypertrophy. We performed immunohistochemical analysis and quantitative RT-PCR on 241 cardiac tissue samples from these cohorts in order to measure protein and mRNA levels of 242 hECSIT. Whilst immunohistochemistry indicated that there was no association between levels 243 of hECSIT protein and a number of echocardiographic parameters, there was a significant 244 negative correlation between hECSIT protein expression in the heart and left ventricular mass 245 index (LVMI) (Figure 3A) suggesting that low cardiac hECSIT protein expression is associated 246 with left ventricular hypertrophy. Furthermore, hECSIT protein also demonstrated negative 247 correlation with indicators of cardiac damage and fibrosis such as Masson’s trichrome staining 248 for collagen volume (Figure 3B, Supplemental Figure 6). In addition cardiac levels of hECSIT 249 protein were inversely related to serum markers of collagen I turnover denoted by the ratio of 250 carboxy- and aminoterminal propeptides of procollagen (PICP and PINP; markers of collagen 251 biosynthesis):carboxyterminal telopeptide of type I collagen (CITP; marker of collagen 252 breakdown) (Figure 3C). Collagen I is the predominant form of collagen in the human heart 253 and these data are consistent with low levels of cardiac hECSIT expression being associated 254 with accumulation of collagen and cardiac fibrosis. We also explored association of cardiac 255 levels of mRNA encoding hECSIT and indicators of fibrosis. hECSIT mRNA was negatively 256 associated with collagen volume as revealed by Picrosirius red staining (Figure 3D). This is 257 also consistent with hECSIT mRNA inversely correlating with mRNA levels of genes encoding 258 the pro-collagen processing enzymes procollagen C-proteinase enhancer (PCPE) and 259 procollagen C-proteinase (PCP) (Figure 3E) and lysyl oxidase that catalyses crosslinking of 260 collagen molecules (Figure 3F). These data are consistent with high level expression of ECSIT

261 being important for cardiac function and protection with low level expression being associated 262 with cardiac hypertrophy and fibrosis.

263 Heart from human ECSIT knock-in mice is characterised by reduced levels of 264 mitochondrial complex I proteins.

265 We next assessed the molecular basis to the cardiac hypertrophy and dysfunction that is 266 associated with low expression of hECSIT. To this end, we utilised an unbiased proteomic 267 approach to systematically and quantitatively analyse dynamic differences between WT and 268 ECSIT+/+ in their heart proteomes at 3-months and 10-months of age. Using liquid 269 chromatography mass spectrometry (LC-MS/MS) label-free quantitation and bioinformatics 270 analysis, 1177 and 1334 proteins were robustly identified in WT versus ECSIT+/+ hearts at 3- 271 months and 10-months respectively (Figure 4A, Supplemental Figure 7A-7B). Restriction of 272 differential expression to ≥2 unique peptide matches and fold changes of ≥1.5 revealed altered 273 expression of 95 proteins when comparing WT with ECSIT+/+ mouse hearts at 3-months of age 274 with 36 proteins showing increased abundance and 59 proteins exhibiting decreased abundance 275 in the ECSIT+/+ hearts (Figure 4B, Supplemental Table 3). At 10 months of age 228 proteins 276 exhibited a significant differential expression pattern with 124 being increased and 104 277 decreased in ECSIT+/+ hearts relative to WT counterparts (Figure 4B, Supplemental Table 3). 278 Interestingly, many of these differentially expressed proteins at both ages were mitochondrial 279 proteins, with functional roles in oxidative phosphorylation (Figure 4A and 4C). Clustering 280 analysis, using the STRING database of known and predicted protein-interaction networks, 281 was performed to investigate if the differentially expressed proteins clustered around molecular 282 and functional networks. Interestingly, ECSIT+/+ heart samples demonstrated increased 283 expression of 26 and decreased expression of 34 mitochondrial-related proteins at 3-months of 284 age. Whilst there was a lack of functional clustering of proteins of increased expression 285 (beyond their mitochondrial localisation), the profile of proteins of reduced expression was 286 intensely clustered around complex I of the mitochondrial electron transport chain (Figure 4D). 287 At 10-months of age the number of mitochondrial proteins of increased expression in ECSIT+/+ 288 heart increased to 99 including cytochrome C oxidase (Cox) subunits of complex IV and 289 subunits of the ATP synthase. The number of mitochondrial proteins of decreased abundance 290 at 10-months in ECSIT+/+ heart was 40 and these again strongly clustered around complex I 291 (Figure 4D). The population of decreased proteins at 10-months also clustered around a 292 network of contractile-associated proteins such as actin, actin-binding proteins (e.g.

293 tropomyosins) and myosin proteins. The findings from the proteomic analysis of heart tissue 294 were confirmed by Western blotting of cardiac tissue and validating that levels of complex I 295 subunits such as NDUFS3 are greatly decreased, whilst Cox IV and Cox V proteins are slightly 296 increased in ECSIT+/+ heart (Figure 4E). Other mitochondrial proteins such as the outer 297 membrane protein TOM20 are present in comparable amounts in WT and ECSIT+/+ heart. 298 Given that ECSIT+/+ heart displayed a signature proteomic profile of greatly reduced levels of 299 many complex I subunits we next compared the activities of complex I in 7 month old WT 300 and ECSIT+/+ heart using Blue Native polyacrylamide gel electrophoresis (BN-PAGE) in 301 conjunction with in-gel activity assay and a specific complex I substrate. The findings clearly 302 demonstrate that heart tissue from 7 month old ECSIT+/+ mice has greatly impaired activity of 303 the individual complex I as well as marked reductions in activity of complex I as part of higher 304 order supercomplexes (Figure 4F). Interestingly such impaired complex I activity is also 305 observed in younger 10 week old ECSIT+/+ mice (Figure 4G) suggesting that complex I 306 deficiency may precede the development of cardiac hypertrophy. In contrast, equivalent 307 analysis demonstrates the same level of activity of complex IV as an individual complex and 308 as part of a supercomplex with complex III in WT and ECSIT+/+ heart suggesting that reduced 309 levels of ECSIT specifically impair the function of complex I (Figure 4F and 4G). These data 310 provide a valuable insight into the molecular basis underlying the cardiac hypertrophy observed 311 in ECSIT+/+ mice. In young mice at 3-months old, the cohort of proteins of decreased expression 312 are mostly subunits of complex I with the primary defect in ECSIT+/+ heart in early life being 313 reduced complex I activity. This will lead to a compromise of mitochondrial and ATP 314 production that will likely lead to adaptive and compensatory response such as hypertrophy. 315 The defect in complex I continues to be observed at 10-months of age but the population of 316 proteins of decreased expression expands to include contractile-associated proteins that play 317 fundamental roles in cardiac muscle contraction and in conjunction with increased fibrosis, is 318 likely a prelude to cardiac failure and premature death.

319

320 Heart from human ECSIT knock-in mice has fragmented mitochondria characterised 321 by greater fission and reduced fusion

322 Electron microscopy was next performed on tissue from the left ventricles of WT and ECSIT+/+ 323 mice in order to understand the consequences of impaired mitochondrial complex I at the 324 ultrastructural level. Tissue from WT mice showed mitochondria in the characteristic highly

325 ordered state along the sarcomeres whereas the arrangement of mitochondria in tissue from 326 ECSIT+/+ mice was less ordered and more irregular (Figure 5A). Such disorder was apparent in 327 heart tissue from male and female mice. Mitochondria from ECSIT+/+ tissue also showed 328 distorted cristae with some mitochondria exhibiting total or partial absence of cristae 329 (highlighted by red arrow in Figure 5A). We also noted that the size of mitochondria from 330 ECSIT+/+ tissue was more heterogeneous than those from WT mice with some of mitochondrial 331 appearing much smaller and rounded than the rest of the population (highlighted by blue arrows 332 in Figure 5A). Quantitative analysis of multiple images confirmed that the mean lengths of 333 both the long and short axes of mitochondria are less in ECSIT+/+ tissue (Figure 5B). The 334 presence of smaller mitochondria in ECSIT+/+ heart suggested potential differences in 335 mitochondrial dynamics between WT and ECSIT+/+ mice. The quality of mitochondria is 336 generally controlled by a dynamic process in which asymmetric fission, mediated by Dynamin 337 related protein (Drp)-1 and mitochondrial fission 1 protein (Fis1), separates a damaged 338 mitochondrion into a depolarized daughter organelle that can be targeted for removal by 339 mitophagy (25) with the other daughter component fusing to the functional mitochondrial pool 340 by the action of the fusion proteins mitofusin 1 and mitofusin 2 in the outer mitochondrial 341 membrane and the fusion protein optic atrophy 1 (Opa 1) in the inner membrane (26). Western 342 blotting of cardiac tissue from WT and ECSIT+/+ mice showed that levels of Fis 1 and 343 phosphorylated Drp-1 were increased in ECSIT+/+ tissue whereas levels of the fusion proteins 344 mitofusin 1 and 2 were reduced with concomitant processing of Opa-1 (Figure 5C-5D) all 345 consistent with ECSIT+/+ tissue showing greater fission and impaired fusion. Since fission 346 generates fragmented mitochondria that can be removed be mitophagy we were keen to 347 compare levels of mitophagy markers in WT and ECSIT+/+ tissue. ECSIT+/+ tissue showed 348 slightly augmented levels of p62 and Parkin but no PINK accumulation (Figure 5C-5D) 349 suggestive of some interference or saturation of the mitophagy process that would normally 350 promote their lysosomal-mediated degradation. Parkin-independent mitophagy mediated by 351 BNIP3L and FUNDC1 receptors was comparable in WT and ECSIT+/+ heart tissue (Figure 5C- 352 5D). The enhanced mitochondrial fission without concomitant increase in mitophagy observed 353 in ECSIT+/+ heart tissue is fully consistent with our ultrastructural electron microscopy analysis 354 showing the accumulation of smaller fragmented mitochondria in these hearts. Whilst 355 ECSIT+/+ heart tissue is characterised by the presence of some fragmented mitochondria the 356 overall mitochondrial content, at least as measured by mitochondrial genome markers, is 357 comparable with WT heart tissue (Figure 5E). However overall mitochondria function is likely

358 compromised given that fragmented mitochondria are frequently damaged and as well as 359 impairing function may also contribute to disease by production of reactive oxygen species.

360 Suppressed expression of hECSIT in cardiomyocytes impairs complex I activity and 361 leads to cellular hypertrophy.

362 Given that hECSIT expression is suppressed in a number of tissues from ECSIT+/+ mice its 363 direct role in cardiomyocytes was next examined in order to assess if its cell-intrinsic function 364 in cardiac cells may underlie the mitochondrial dysfunction and cardiac hypertrophy observed 365 in ECSIT+/+ mice. We were also keen to show that the latter effects in ECSIT+/+ mice were due 366 to reduced levels of hECSIT and not due to differences in function for hECSIT and mEcsit. To 367 this end, we used hECSIT-specific siRNA to suppress endogenous expression of hECSIT in 368 human AC16 cardiomyocytes (Figure 6A). Immunofluorescent staining of hECSIT 369 knockdown cells showed an apparent increase in the overall cell size (Figure 6B) and this was 370 further confirmed by quantitative analysis of cell surface area (Figure 6C). Such hypertrophy 371 was reversible since it was not apparent when hECSIT-specific siRNA knockdown cells were 372 grown for an additional 7 days to recover from transient siRNA-mediated downregulation of 373 hECSIT (Figure. 6D). The magnitude of cell hypertrophy induced by knockdown of hECSIT 374 was comparable to that observed when cells were treated with Angiotensin II, a well known 375 driver of cardiac hypertrophy (Figure 6C). Thus reduced expression of hECSIT in human 376 cardiomyocytes that leads to cellular hypertrophy mirrors at a cell level the cardiac 377 hypertrophy observed in the ECSIT+/+ mice that expresses low levels of hECSIT. Furthermore, 378 suppression of hECSIT in cardiomyocytes also shows the same effects on mitochondrial 379 function as described in cardiac tissue from ECSIT+/+ mice in that BN-PAGE analysis revealed 380 that knockdown of hECSIT in AC16 cells impaired the activity of complex I as part of 381 mitochondrial supercomplexes whereas other complexes such as complex IV were unaffected 382 (Figure 6E). Such impaired complex I activity in hECSIT knockdown AC16 cells is also 383 consistent with reduced levels of complex I subunits, as typified by NDUFS3, whereas complex 384 IV protein and other mitochondrial proteins like Tom 20 were not affected by suppression of 385 hECSIT (Figure 6F). Knockdown of hECSIT also resulted in modest increases in levels of p62 386 and PINK1. Similar to ECSIT+/+ cardiac tissue, hECSIT knockdown AC16 cells showed signs 387 of increased mitochondrial fission and reduced fusion as indicated by enhanced levels of the 388 fission protein Fis1 and increased processing of the fusion protein (Opa1) (Figure 6F). Notably 389 the levels of mitofusin 1 and 2 were not affected by knockdown of hECSIT suggesting that

390 additional factors may be required to manifest their reduced expression in ECSIT+/+ cardiac 391 tissue. These data show that threshold levels of hECSIT are required for normal mitochondrial 392 function and dynamics in cardiomyocytes with reduction below these levels resulting in 393 compensatory changes such as cellular hypertrophy. This closely mirrors the cardiac 394 pathophysiology observed in ECSIT+/+ mice that express low amounts of hECSIT.

395 Reduced levels of hECSIT promotes a metabolic shift from oxidative phosphorylation to 396 glycolysis.

397 Given that the knockdown of hECSIT in AC16 cardiomyocytes models many of the 398 pathophysiological features in hearts from ECSIT+/+ mice, this model was used to better 399 understand how the role of hECSIT in mitochondrial function contributes to cardiac 400 physiology. Since cardiac hypertrophy is frequently a response to a deficit in serving the 401 enormous bioenergetics requirements of the heart and mitochondria are the major producers of 402 ATP in cardiomyocytes to satisfy this energy demand, we next assessed the role of hECSIT in 403 primary metabolism and respiration in cardiomyocytes. To this end, the effect of hECSIT 404 knockdown in AC16 cardiomyocytes on metabolic flux analysis was investigated. Oxygen 405 consumption rate (OCR) was measured to characterise the effect of hECSIT knockdown on the 406 various contributory pathways to respiration. Whilst basal respiration, maximal respiration and 407 spare respiratory capacity were the same in AC16 cells transfected with control or hECSIT- 408 specific siRNA, the OCR that was associated with ATP production was significantly 409 diminished in hECSIT knockdown cells (Figure 7A). Maximal respiration was not affected 410 since the shortfall in respiration associated with ATP production was compensated by increased 411 respiration due to proton leak. The decreased respiration associated with ATP production 412 translates to reduced levels of oxidative phosphorylation and metabolic flux analysis was next 413 applied to assess if hECSIT knockdown cells demonstrated a re-programming to the glycolytic 414 pathway. The latter was characterised by measuring the extracellular acidification rates 415 (ECAR) in control and hECSIT knockdown cells. AC16 cells, transfected with hECSIT- 416 specific siRNA, demonstrated a tendency to an increase ECAR in glycolysis (Figure 7B) with 417 a more precise measurement of glycolytic rate confirming that hECSIT knockdown results in 418 increased rates of basal glycolysis (Figure 7C). The overall glycolytic capacity of the cells was 419 not affected and as expected the increased rate of glycolysis was associated with a decrease in 420 the glycolytic reserve (Figure 7B). These data clearly indicate that reduced expression of 421 hECSIT in cardiomyocytes results in reduced levels of oxidative phosphorylation with a

422 metabolic re-programming to increased rates of glycolysis. Metabolic flux also allowed for 423 calculation of the amounts of ATP generated by these pathways and confirmed that hECSIT 424 knockdown cells manifested higher rates of ATP production by glycolysis and less ATP by 425 oxidative phosphorylation than control cells (Figure 7D, left panel). Indeed the magnitude of 426 this shift is highlighted by the ratio of ATP produced by oxidative phosphorylation:ATP 427 produced by glycolysis decreasing from 1.45 in control cells to 0.35 in hECSIT knockdown 428 cells (Figure 7D, right panel). These findings are consistent with a model in which reduced 429 levels of hECSIT in knockdown cells leads to impairment of complex I activity, reduced 430 oxidative phosphorylation and ATP production resulting in a compensatory shift to ATP 431 production by increased rates of glycolysis. The pathophysiological reliance of this metabolic 432 programming in response to reduced levels of ECSIT is supported by similar metabolic patterns 433 in isolated mitochondria from ECSIT+/+ hearts. Thus ECSIT+/+ mitochondria showed reduced 434 oxygen consumption (Figure 8A),impairment of basal and maximal respiration (Figure 8B) 435 and reduced rates of ATP production (Figure 8B) relative to isolated mitochondrial from WT 436 mice. In addition cardiac tissue from ECSIT+/+ mice showed reduced levels of ATP (Figure 437 8C). This reflects a re-programming of primary metabolism to increased glycolysis that is 438 confirmed by increased levels of the glycolytic end product lactic acid in cardiac tissue from 439 ECSIT+/+ mice (Figure 8D). The reduction in oxidative phosphorylation is also associated with 440 incomplete reduction of oxygen to form ROS products as measured by increased levels of

441 H2O2 (Figure 8E). The compensatory shift to glycolysis clearly does not fully restore the ATP- 442 generating capacity of cardiomyocytes and in the in vivo setting this will likely lead to an 443 adaptive cardiac hypertrophy response that we observe in ECSIT+/+ mice.

444 All of these data show that the levels of hECSIT is a critical determinant of mitochondrial 445 function and we now provide the first insight into the key regulatory processes that dictate the 446 stability of this protein.

447

448

449 Discussion 450 Prior to this study, a number of functions had already been ascribed to Ecsit proteins including 451 an indispensable role of mEcsit in BMP signalling in early development, as a positive regulator 452 of the NFB pathway and inflammatory gene expression, and as an important factor in 453 promoting assembly of complex I in the electron transport chain of the mitochondria (27-30). 454 Furthermore, mEcsit was also reported to facilitate the juxtaposition of phagosomes and 455 mitochondria to enhance mROS production and bactericidal activity in response to Salmonella 456 infection (31). Exploiting the labile nature of hECSIT, we now show that whereas early 457 developmental pathways, activation of NFB and induction of pro-inflammatory cytokines are 458 all functionally intact with even low levels of hECSIT, the mitochondrial functions of the latter 459 are critically dependent on its expression levels. In the process we also develop a new mouse 460 model that reveals the first physiological role for ECSIT in an in vivo setting. We describe the 461 critical contribution of hECSIT to cardiac physiology with threshold levels of its expression 462 being required to ensure normal cardiac function. We show that sub-threshold levels of hECSIT 463 contribute to cardiac disease as revealed by the manifestation of cardiac hypertrophy in 464 ECSIT+/+ mice and the association of low levels of hECSIT with myocardial hypertrophy and 465 cardiac fibrosis in human cohorts. It should be noted that due to lack of availability of human 466 tissue from the left ventricle, fibrosis measurement were performed on atrial appendages that 467 were available from clinically annotated cohorts. Whilst hECSIT plays a fundamental role in 468 mitochondrial function, it is interesting to note that low levels of hECSIT primarily manifest 469 in cardiac dysfunction leading to cardiac hypertrophy without showing any other obvious 470 effects on the other main organs. This may be due to the extremely high demands of heart 471 requiring highly efficient electron transport and oxidative phosphorylation for ATP production 472 with any deficiencies in these processes, such as impaired complex I under conditions of low 473 hECSIT expression, resulting in essential compensatory processes and adaptation as occurs in 474 cardiac hypertrophy. The lower energy demands of other tissues may be more tolerant of loss 475 of complex I and mitochondrial function. Given that expression levels of hECSIT appears to 476 be critical for cardiac physiology, it is interesting to speculate on why evolutionary pressure 477 would not select against hECSIT losing its protein stability whereas in other mammals such as 478 mice, mECSIT is significantly more stable. There are significant differences in cardiac 479 physiology in mouse and man. The high workload of the heart places major metabolic and 480 energy demands on this organ. This is especially relevant to mice that have average heartbeats 481 of 600/minute whereas the average heartbeat in humans is less by an order of magnitude. Thus,

482 there is strong pressure in mouse to maximise mitochondrial output whereas this need also 483 applies in human heart but may not be subject to the extreme demands of mouse heart. Indeed 484 whilst the stability of hECSIT is adequate for human demands, the phenotype of cardiac 485 hypertrophy in humanised hECSIT knock-in mice demonstrates that its stability is not 486 sufficient to cater for the energy demands in mouse, thus promoting compensatory and adaptive 487 processes such as hypertrophy. A recent report has also highlighted a potential evolutionary 488 pressure that may favourably select the labile nature of hECSIT. Thus a recurrent and 489 hyperactive ECSIT mutation, that results in greatly augmented downstream activation of 490 NADPH and NFB and induction of pro-inflammatory cytokines, can trigger the fatal 491 hyperinflammatory disease hemophagocytic syndrome (32). The labile nature of hECSIT is 492 likely to exercise tighter control over its downstream inflammatory effects and so minimise the 493 risk of hyperinflammation. Notably, embryonic lethality, caused by mEcsit deficiency, is 494 rescued by knock-in of hECSIT indicating that the human orthologue can serve the same 495 developmental role in the BMP pathway. Interestingly no hECSIT mutations have been 496 reported to date and this may reflect that total loss of function is incompatible with life. 497 However, mutations in another factor, ACAD9, which also associates with NDUFAF1 and 498 ECSIT, appear to be a relatively common cause of complex I deficiency presenting as 499 cardiomyopathy and/or exercise intolerance (10) .

500 We also map out the cellular and molecular basis to the critical role of hECSIT in normal 501 cardiac function. Thus, hECSIT is essential for assembly and maintenance of mitochondrial 502 complex I and driving efficient oxidative phosphorylation and ATP production in 503 cardiomyocytes. Low level expression of hECSIT causes reduction in oxidative 504 phosphorylation and this likely leads to dysfunctional and depolarised mitochondria. Indeed 505 the heart tissue from ECSIT+/+ mice show some fragmented mitochondria that is associated 506 with increased fission and reduced fusion. Normally fission is a prologue to removal of 507 damaged mitochondria by mitophagy but instead mitophagy is not effectively or sufficiently 508 activated in ECSIT+/+ mice and this likely leads to the accumulation of damaged fragmented 509 mitochondria that is observed by electron microscopy. Limited mitophagy may be due to the 510 reduced levels of mitofusin 2, a receptor for Parkin or may be a more direct consequence of 511 reduced levels of hECSIT in light of a recent report describing a role for mEcsit in mitophagy 512 by acting as a substrate for Parkin (14). In response to reduced oxidative phosphorylation under 513 conditions of low hECSIT, there is a metabolic shift to glycolysis to cater for reduced 514 production of ATP. However, this switch does not fully compensate for the reduced ATP

515 production and this likely underlies the ensuing hypertrophy that is likely to be adaptive at the 516 early stage but over time progresses to a pathologic situation and heart failure.

517 There is a close association of mitochondrial dysfunction with cardiac disease but the molecular 518 basis to this relationship is not fully delineated. Given that deficiencies in complex I is a very 519 common defect in mitochondrial diseases including heart failure and cardiomyopathy there is 520 much interest in gaining a better understanding of disease progression and new treatments. The 521 lack of good models has contributed to the slow progress in this regard. We have now 522 developed a new model that may be very valuable in dissecting the pathophysiology of cardiac 523 disease progression that is associated with complex I and mitochondrial dysfunction and in 524 evaluating the preclinical therapeutic potential of new lead candidates for disease treatment. 525 The models offers a number of attractive features. Firstly, the cardiac hypertrophy associated 526 with the model develops gradually over time and this may have application in modelling the 527 slow progression of human cardiac dysfunction with age. This may complement and indeed 528 surpass many of the widely used mouse models that tend to use relatively acute pressure 529 overload to induce cardiac pathology but may not capture the mechanisms underlying the more 530 gradual pathogenesis in humans. Interestingly, a recent study has reported the expression of 531 hECSIT to decrease with age (33) and in the context of our present findings this may have 532 important consequences for cardiac disease as part of the ageing process. Other genetic models 533 have also been generated to study cardiac hypertrophy and disease. These include a transgenic 534 mouse overexpressing myotropin, that stimulates myocyte growth (34) and a knockout mouse 535 lacking cardiac myosin binding protein-C (cMyBP-C) (35). Whilst these models have been 536 valuable in terms of studying cardiac disease, the interpretation of data from such mice is 537 complicated by cardiac disease developing very early with the cMyBP-C knockout showing 538 cardiac hypertrophy and remodelling during perinatal development. The slow and gradual 539 development of cardiac hypertrophy in the ECSIT+/+ mice offers much promise to address this 540 current limitation in animal models to study cardiac disease progression. Furthermore, the 541 reduced levels of most of the protein subunits in mitochondrial complex I in ECSIT+/+ mice 542 offers opportunity to better understand the role of complex I in a number of diseases. 543 Deficiencies in mitochondrial complex I in man manifest as diseases ranging from hypertrophic 544 cardiomyopathy to liver disease and neuropathological conditions, including Parkinson’s 545 disease. There is a paucity of genetic models of complex I deficiency that allow for study of 546 slow disease progression since deficiency in complex I subunits tends to lead to developmental 547 lethality or death at very young ages. Here, we generate humanised hECSIT mice with complex

548 I deficiency but with a life span of ~ 1 year. These mice may serve as a valuable resource in 549 future study of complex I-related diseases.

550 In summary these studies describe the first physiological role for hECSIT in cardioprotection 551 and in the process develops a model to study disease progression in mitochondria dysfunction 552 and heart disease.

553

554 Methods

555 Mice

556 Humanized mice were generated by Taconic Artemis using proprietary technology. To 557 generate human ECSIT knock-in mice, heterozygotes (ECSIT+/-) for the targeted allele were 558 bred with mice containing a Flpe transgene (C57BL/6N-Tg(CAG-Flpe)2Arte). This resulted 559 in the replacement of exons 3-9 of murine ECSIT with exons 2-8 of human ECSIT. 560 The Flpe transgene was removed by breeding the resulting mice with C57BL/6N mice during 561 colony expansion. Mice were genotyped by PCR analysis of DNA isolated from ear punches 562 using primers 'a', ‘b’. All animals were used at age 8-12 weeks unless indicated otherwise. 563 Animals were allocated to experimental groups based on genotype, age and sex. No 564 randomization methods were used.

565 Mitochondria isolation

566 Mouse hearts (two pairs 7 months old, gender mixed) or cells (AC16) were washed with ice- 567 cold washing buffer (10 mM HEPES pH 7.2 containing 0.3 M Sucrose,0.2 mM EDTA), 568 minced, and homogenized in Isolation buffer (10 mM HEPES pH 7.4 containing 0.3 M 569 Sucrose, 0.2 mM EDTA, 1 mg/ml BSA, 1× trypsin-EDTA solution). After the homogenate was 570 centrifuged at 4°C for 15 min at 600 g, the supernatant was collected and further centrifuged 571 at 4°C for 15 min at 8500 g. The pellet was re-suspended for functional assessment. The protein 572 concentration of the mitochondrial preparation was determined by BCA assay.

573 Blue Native Polyacrylamide Gel Electrophoresis (BN-PAGE)

574 Blue Native-PAGE was conducted using the NativePAGETM system (Invitrogen). Briefly, 575 isolated cardiac mitochondria (50 µg) were solubilized in 20 µl of NativePAGE sample buffer 576 supplemented with 2% (w/v) digitonin for 20 min on ice, then centrifuged at 20,000 g at 4°C 577 for 10 min. After centrifugation, the supernatants were collected. Coomassie blue G-250 578 (Invitrogen) was added to the supernatant to obtain a dye/detergent mass ratio of 4/1 and then 579 the protein was loaded into a NativePAGE 3-12% gradient gel (Invitrogen). The gel was run at 580 150 V for 30 min, and then the light blue buffer in the inner chamber was refreshed with 200 581 ml of cathode buffer. The gel was run at 250 V for 150 min.

582 For in-gel assay of complex I/IV activities, gels were incubated in 10 ml of Complex I substrate 583 solution (2 mM Tris-HCl pH 7,4 containing 0.1 mg/ml NADH, 2.5 mg/ml Nitrotetrazolium

584 Blue chloride) or Complex IV substrate solution (0.5 mg/ml Diaminobenzidine (DAB) and 1 585 mg/ml cytochrome c in 45 mM phosphate buffer pH 7.4). Complex I and Complex IV activities 586 manifested as violet and brown bands respectively. Reactions were terminated with 10% (v/v) 587 acetic acid followed by washing with water and images captured by scanning.

588 Lactate assay

589 Left ventricular tissue from 6-7 months mice (n=4, two pairs male and two pairs female) was 590 homogenized in RIPA buffer (50 mM HEPES pH 7.5, 10% (v/v) glycerol, 0.5% (w/v) CHAPS, 591 0.5% (v/v) Triton-X-100, 150 mM NaCl, 1 mM Na3VO4, 1 mM EDTA, 1 mM PMSF and 592 complete protease inhibitor mixture). Protein concentration was determined by BSA assay, 593 Perchloric acid (PCA) was added to a final concentration of 1M to precipitate protein. 594 Supernatants were neutralised by adding ice-cold 2 M KOH to reach pH 6.5-8.0. L-lactate was 595 measured using L-lactate assay Kit (MAK064-1KT).

596 Mitochondrial mass

597 For mitochondrial mass determination, genomic DNA was extracted from tissues (n=3, around 598 7 months old, male) using the DirectPCR Lysis reagent (102-T), following the manufacturer’s 599 instructions. DNA concentrations were determined by Nanodrop. 10ng DNA were used to 600 perform qPCR for mitochondrial gene mtCOI and nuclear gene ndufv1. Ratios of 2-ΔCT for

601 mtCOI over ndufv1 were averaged and fold of WT is shown. The primers used see S4 Table.

602

603 RNA interference

604 Human specific ECSIT (4390825, Assay ID s224197) and control siRNAs (4390844) were 605 purchased from ThermoFisher. siRNA (40 nM) was delivered to AC16 via transfection with 606 Lipofectamine RNAiMAX Transfection Reagent (13778030) according to the manufacturer’s 607 instructions and allowed to recover for 48h prior to experiments.

608 Metabolic Flux analysis of cells

609 Cell metabolism was measured using a Seahorse XF96 Extracellular Flux Analyzer according 610 to the manufacturer's instructions. Briefly, AC16 cells (pre-treated with control or hECSIT 611 siRNA) were seeded onto an XF96 microplate at 50,000 cells/well. The cellular oxygen 612 consumption rate (OCR) used the in mitochondria stress test was monitored in seahorse assay

613 medium (103575-100) supplemented with 2 mM glutamine, 1 mM sodium pyruvate, and 25 614 mM glucose (pH 7.4 at 37 °C), following the sequential addition of oligomycin (2 μM), 615 carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP, 2 μM), and rotenone (1 μM) 616 and antimycin A (AA, 0.5 μM). Extracellular acidification rate (ECAR) used in the glycolysis 617 stress test was monitored in seahorse assay medium (103575-100) supplemented with 2 mM 618 glutamine, (pH 7.4 at 37 °C), following the sequential addition of glucose (10 mM), oligomycin 619 (2 μM), 2-DG (50 mM). Glycolysis proton efflux rates (glycoPER) used in the glycolytic rate 620 was monitored in seahorse assay medium (103575-100) supplemented with 2 mM glutamine, 621 10 mM glucose, 1mM pyruvate, (pH 7.4 at 37 °C), following the sequential addition of 622 rotenone (1 μM) and antimycin A (AA, 0.5 μM), and 2-DG (50 mM). Real time ATP rate was 623 monitored in seahorse assay medium (103575-100) supplemented with 2 mM glutamine, 10 624 mM glucose, 1mM pyruvate, (pH 7.4 at 37 °C), following the sequential addition of oligomycin 625 (2 μM) rotenone (1 μM) and antimycin A (AA, 0.5 μM).

626 Metabolic Flux analysis of isolated mitochondria

627 Mitochondrial bioenergetics in isolated mitochondria from hearts was determined as described 628 previously (36). Hearts from 10-week-old mice were rinsed and isolated in MIB1 buffer (210 629 mM d-Mannitol, 70 mM sucrose, 5mM HEPES, 1mM EGTA and 0.5% fatty acid-free BSA, 630 pH 7.2). Isolated mitochondrial was seeded onto an XF96 microplate at 6 µg/well. 631 Mitochondrial assay was performed in MAS1 buffer (220 mM d-Mannitol, 70 mM sucrose, 10

632 mM KH2PO4, 5 mM MgCl2, 2 mM HEPES, 1 mM EGTA and 0.2% fatty acid-free BSA, pH 633 7.2 at 37 °C) supplemented with 10 mM Glutamate and 5 mM Malate for complex I-driven 634 respiration in a Seahorse XF96 Extracellular Flux Analyzer. Compounds (4 mM ADP, 2.5 635 µg/ml oligomycin, 4 µM FCCP and 2 µM rotenone) were injected to each well in sequence as 636 running procedure described previously (36).

637 ATP measurement from heart tissue

638 ATP production was measured by the ATP assay (MAK190-1KT) according to the 639 manufacturer’s instructions. Briefly, left ventricle tissue was homogenised into ATP assay 640 buffer. Protein concentration of clarified supernatants was determined by BCA assay for further 641 normalization. Clarified supernatants were deproteinized through 10 kDa molecular weight 642 cut-off spin filter (MRCPRT010). Deproteinized homogenate or ATP standards were incubated 643 with ATP probe and converter as standard protocol at room temperature for 30 mins, protected

644 from light. Fluorescence was measured with an excitation wavelength of 530 nm and 645 fluorescence emission detection at 620 nm.

646 Amplex Red-based Hydrogen peroxide measurement in heart tissue

647 Hydrogen peroxide was measured by the Amplex Red hydrogen peroxide assay (A22188) 648 according to the manufacturer’s instructions. Briefly, left ventricle was homogenised into 649 Amplex Red reaction buffer. Clarified supernatants or H2O2 standards were incubated with 650 100 µM Amples Red and 0.2 U/ml HRP at room temperature for 30 mins, protected from light. 651 Fluorescence was measured with an excitation wavelength of 530 nm and fluorescence 652 emission detection at 620 nm. BCA was performed to determine total protein for further 653 normalization.

654 Transmission Electron Microscopy

655 Pairs of 6-month old females and 7-month old males were sacrificed and slices (1-2 mm) of 656 left ventricular tissue were fixed in Karnovsky solution (Thermo Fisher, 50-980-497) overnight 657 at 4 °C. Specimens were then washed in PBS before being immersion fixed in 2.5% 658 glutaraldehyde and 2% paraformaldehyde in PBS overnight at 4oC. Fixed samples were

659 washed 3x in PBS then post-fixed in aqueous 1% OsO4, 1% K3Fe(CN)6 for 1 hour. Following 660 3 PBS washes, the pellet was dehydrated through a graded series of 30-100% ethanol, 100% 661 propylene oxide then infiltrated in 1:1 mixture of propylene oxide:Polybed 812 epoxy resin 662 (Polysciences, Warrington, PA) for 1 hr. After several changes of 100% resin over 24 hrs, the 663 samples are embedded in molds, cured at 37oC overnight, followed by additional hardening at 664 65oC for two more days. Semi-thin (300 nm) sections are heated onto glass slides, stained with 665 1% toluidine blue and imaged using light microscopy to assure proper tissue orientation. 666 Ultrathin (60-70 nm) sections were collected on 100 mesh copper grids, stained with 4% uranyl 667 acetate for 10 minutes, followed by 1% lead citrate for 7 minutes. Sections were imaged using 668 a JEOL JEM 1011 or JEOL JEM 1400 transmission electron microscope (Peabody, MA) at 80 669 kV fitted with a side mount AMT digital camera (Advanced Microscopy Techniques, Danvers, 670 MA). The short axis and long axis were calculated according 190 mitochondria from 14 671 individual images (7 from 6-month female and the other 7 from 7-month male). 672

673 Echocardiography

674 6-7 months old mice (n=8, male) were anaesthetised with 1.5% isofluorane/oxygen, placed on 675 a warming pad, and imaged in the supine position using a Vevo770 ultrasound system with 676 high-frequency 45MHz RMV707B scanhead (VisualSonics, Inc.). M-mode parasternal short- 677 axis scans at papillary muscle level were used to quantify LV end-diastolic (LVEDD) and end- 678 systolic diameters (LVESD) from which percentage fractional shortening was calculated 679 (LVEDD−LVESD)/LVEDD*100. Pulse-wave Doppler was used to assess mitral valve flow 680 (E/A ratio), LV isovolumetric relaxation time (IVRT) and myocardial performance index 681 (MPI), as reliable measures of diastolic function. Electrocardiogram (ECG) and heart rate 682 recordings were made by attachment of mouse paws to electrodes on the system platform.

683 Cell surface area

684 AC16 cells (pre-treated with control or hECSIT siRNA) were cultured in DMEM/12 overnight 685 and treated with angiotensin II (Ang II) (2 M) for 24 hours. Cells were fixed with 4% (w/v) 686 paraformaldehyde for 15 minutes, permeabilized with 0.1% (v/v) Triton X-100 in PBS for 10 687 minutes, blocked with a 3% (w/v) BSA solution for 1 h and incubated with anti--actinin (05- 688 384) antibody overnight at 4°C. Immunofluorescence was analysed with a laser scanning 689 confocal microscopy and the surface areas were measured using ImageJ software. 100 cells for 690 each condition were randomly measured from three individual biological replicates. In some 691 experiments, knockdown cells were allowed to recover for 7 days in DMEM/12 before 692 repeating the procedures above.

693 Statistical analysis

694 Statistical analysis was performed using an unpaired 2-tailed Student’s t-test or one-way 695 ANOVA as specified for multiple comparison between groups. Data are presented as mean 696 ± SD unless otherwise indicated. P≤0.05 was considered statistically significant.

697 Study approval

698 All animal experiments were performed in accordance with the regulations and guidelines of 699 the Health Products Regulatory Authority (HPRA) and protocols approved by the Research 700 Ethics committee of Maynooth University. The study utilized a cohort of 38 patients 701 undergoing elective cardiac surgery for coronary artery by-pass grafting or valve replacement 702 at the Cardiology Department of the Blackrock Clinic in Dublin, Ireland. Right atrial 703 appendage tissue and peripheral venous blood were collected from this cohort. All subjects

704 gave written informed consent to participate in this study. The Ethics Committee at St. 705 Vincent’s University Hospital approved all study protocols which conformed to the principles 706 of the Helsinki Declaration.

707

708 709

710

711 Author Contributions 712 L.X developed the concept, designed and performed experiments, analyzed data, prepared the 713 figures and edited the manuscript; F.H. developed the concept, designed and performed 714 experiments. N.D. and B.W. performed experiments and analyzed data; A.H. designed, 715 performed and analysed proteomic studies and performed bioinformatics and 716 immunofluorescent studies; K.S.E and D.J.G. designed, performed, analyzed and supervised 717 the Echocardiography studies; D.B.S. performed the electron microscopy studies. E.O. 718 performed the flow cytometry analysis. A.M.R. and R.I. performed S. Typhimurium infections. 719 J.M.R. performed the confocal microscopy analysis. N.G., C.J.W., K.M. and M.T.L. designed 720 and analysed the ex-vivo clinical study, performed correlation analysis of hECSIT expression 721 and immunohistochemical, biochemical, clinical and imaging parameters and edited the 722 manuscript. P.N.M. conceived the study, supervised the overall project, analyzed data and 723 wrote the manuscript.

724 725 Acknowledgments

726 This publication has emanated from research conducted with the financial support of Science 727 Foundation Ireland (SFI) under Grant Numbers SFI/16/IA/4622. The analysis of ex-vivo 728 human tissue samples was part funded by a Health Research Board award 729 (HRAPOR/2012/119).

730

731

732 References 733 1. Humphries F, and Moynagh PN. Molecular and physiological roles of Pellino E3 ubiquitin 734 ligases in immunity. Immunol Rev. 2015;266(1):93-108. 735 2. Chen ZJ. Ubiquitination in signaling to and activation of IKK. Immunol Rev. 2012;246(1):95- 736 106. 737 3. Chen ZJ. Ubiquitin signalling in the NF-kappaB pathway. Nat Cell Biol. 2005;7(8):758-65. 738 4. Kopp E, Medzhitov R, Carothers J, Xiao C, Douglas I, Janeway CA, and Ghosh S. ECSIT is 739 an evolutionarily conserved intermediate in the Toll/IL-1 signal transduction pathway. Genes 740 Dev. 1999;13(16):2059-71. 741 5. Wi SM, Moon G, Kim J, Kim ST, Shim JH, Chun E, and Lee KY. TAK1-ECSIT-TRAF6 742 complex plays a key role in the TLR4 signal to activate NF-kappaB. J Biol Chem. 743 2014;289(51):35205-14. 744 6. Mi Wi S, Park J, Shim JH, Chun E, and Lee KY. Ubiquitination of ECSIT is crucial for the 745 activation of p65/p50 NF-kappaBs in Toll-like receptor 4 signaling. Mol Biol Cell. 746 2015;26(1):151-60. 747 7. Wen H, Ma H, Cai Q, Lin S, Lei X, He B, Wu S, Wang Z, Gao Y, Liu W, et al. Recurrent 748 ECSIT mutation encoding V140A triggers hyperinflammation and promotes hemophagocytic 749 syndrome in extranodal NK/T cell lymphoma. Nat Med. 2018;24(2):154-64. 750 8. Alston CL, Compton AG, Formosa LE, Strecker V, Olahova M, Haack TB, Smet J, Stouffs K, 751 Diakumis P, Ciara E, et al. Biallelic Mutations in TMEM126B Cause Severe Complex I 752 Deficiency with a Variable Clinical Phenotype. Am J Hum Genet. 2016;99(1):217-27. 753 9. McKenzie M, and Ryan MT. Assembly factors of human mitochondrial complex I and their 754 defects in disease. IUBMB Life. 2010;62(7):497-502. 755 10. Nouws J, Nijtmans L, Houten SM, van den Brand M, Huynen M, Venselaar H, Hoefs S, 756 Gloerich J, Kronick J, Hutchin T, et al. Acyl-CoA dehydrogenase 9 is required for the 757 biogenesis of oxidative phosphorylation complex I. Cell Metab. 2010;12(3):283-94. 758 11. Vogel RO, Janssen RJ, van den Brand MA, Dieteren CE, Verkaart S, Koopman WJ, Willems 759 PH, Pluk W, van den Heuvel LP, Smeitink JA, et al. Cytosolic signaling protein Ecsit also 760 localizes to mitochondria where it interacts with chaperone NDUFAF1 and functions in 761 complex I assembly. Genes Dev. 2007;21(5):615-24. 762 12. West AP, Brodsky IE, Rahner C, Woo DK, Erdjument-Bromage H, Tempst P, Walsh MC, Choi 763 Y, Shadel GS, and Ghosh S. TLR signalling augments macrophage bactericidal activity through 764 mitochondrial ROS. Nature. 2011;472(7344):476-80. 765 13. West AP, Shadel GS, and Ghosh S. Mitochondria in innate immune responses. Nat Rev 766 Immunol. 2011;11(6):389-402. 767 14. Carneiro FRG, Lepelley A, Seeley JJ, Hayden MS, and Ghosh S. An Essential Role for ECSIT 768 in Mitochondrial Complex I Assembly and Mitophagy in Macrophages. Cell Rep. 769 2018;22(10):2654-66. 770 15. Xiao C, Shim JH, Kluppel M, Zhang SS, Dong C, Flavell RA, Fu XY, Wrana JL, Hogan BL, 771 and Ghosh S. Ecsit is required for Bmp signaling and mesoderm formation during mouse 772 embryogenesis. Genes Dev. 2003;17(23):2933-49. 773 16. Brown DA, Perry JB, Allen ME, Sabbah HN, Stauffer BL, Shaikh SR, Cleland JG, Colucci 774 WS, Butler J, Voors AA, et al. Expert consensus document: Mitochondrial function as a 775 therapeutic target in heart failure. Nat Rev Cardiol. 2017;14(4):238-50. 776 17. Scheubel RJ, Tostlebe M, Simm A, Rohrbach S, Prondzinsky R, Gellerich FN, Silber RE, and 777 Holtz J. Dysfunction of mitochondrial respiratory chain complex I in human failing 778 myocardium is not due to disturbed mitochondrial gene expression. J Am Coll Cardiol. 779 2002;40(12):2174-81. 780 18. Marin-Garcia J, Goldenthal MJ, Damle S, Pi Y, and Moe GW. Regional distribution of 781 mitochondrial dysfunction and apoptotic remodeling in pacing-induced heart failure. J Card 782 Fail. 2009;15(8):700-8. 783 19. Mimaki M, Wang X, McKenzie M, Thorburn DR, and Ryan MT. Understanding mitochondrial 784 complex I assembly in health and disease. Biochim Biophys Acta. 2012;1817(6):851-62.

785 20. Su AI, Wiltshire T, Batalov S, Lapp H, Ching KA, Block D, Zhang J, Soden R, Hayakawa M, 786 Kreiman G, et al. A gene atlas of the mouse and human protein-encoding transcriptomes. Proc 787 Natl Acad Sci U S A. 2004;101(16):6062-7. 788 21. Wu C, Jin X, Tsueng G, Afrasiabi C, and Su AI. BioGPS: building your own mash-up of gene 789 annotations and expression profiles. Nucleic Acids Res. 2016;44(D1):D313-6. 790 22. Wu C, Macleod I, and Su AI. BioGPS and MyGene.info: organizing online, gene-centric 791 information. Nucleic Acids Res. 2013;41(Database issue):D561-5. 792 23. Wu C, Orozco C, Boyer J, Leglise M, Goodale J, Batalov S, Hodge CL, Haase J, Janes J, Huss 793 JW, 3rd, et al. BioGPS: an extensible and customizable portal for querying and organizing gene 794 annotation resources. Genome Biol. 2009;10(11):R130. 795 24. Mancia G, Fagard R, Narkiewicz K, Redon J, Zanchetti A, Bohm M, Christiaens T, Cifkova R, 796 De Backer G, Dominiczak A, et al. 2013 ESH/ESC guidelines for the management of arterial 797 hypertension: the Task Force for the Management of Arterial Hypertension of the European 798 Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). Eur Heart 799 J. 2013;34(28):2159-219. 800 25. Twig G, Elorza A, Molina AJ, Mohamed H, Wikstrom JD, Walzer G, Stiles L, Haigh SE, Katz 801 S, Las G, et al. Fission and selective fusion govern mitochondrial segregation and elimination 802 by autophagy. EMBO J. 2008;27(2):433-46. 803 26. Nan J, Zhu W, Rahman MS, Liu M, Li D, Su S, Zhang N, Hu X, Yu H, Gupta MP, et al. 804 Molecular regulation of mitochondrial dynamics in cardiac disease. Biochim Biophys Acta Mol 805 Cell Res. 2017;1864(7):1260-73. 806 27. Xiao C SJ, Klüppel M, Zhang SS, Dong C, Flavell RA, Fu XY, Wrana JL, Hogan BL, Ghosh 807 S. Ecsit is required for Bmp signaling and mesoderm formation during mouse embryogenesis. 808 Genes Dev. 2003;17(23):2933-49. 809 28. Mi Wi S PJ, Shim JH, Chun E, Lee KY. Ubiquitination of ECSIT is crucial for the activation 810 of p65/p50 NF-κBs in Toll-like receptor 4 signaling. Mol Bio Cell. 2013;26(1):151-60. 811 29. Vogel RO JR, van den Brand MA, Dieteren CE, Verkaart S, Koopman WJ, Willems PH, Pluk 812 W, van den Heuvel LP, Smeitink JA, Nijtmans LG. Cytosolic signaling protein Ecsit also 813 localizes to mitochondria where it interacts with chaperone NDUFAF1 and functions in 814 complex I assembly. Genes Dev. 2007;21(5):615-24. 815 30. Wi SM MG, Kim J, Kim ST, Shim JH, Chun E, Lee KY. TAK1-ECSIT-TRAF6 complex plays 816 a key role in the TLR4 signal to activate NF-κB. J Biol Chem. 2014;289(51):35205-14. 817 31. West AP, Brodsky IE, Rahner C, Woo DK, Erdjument-Bromage H, Tempst P, Walsh MC, Choi 818 Y, Shadel GS, and Ghosh S. TLR signalling augments macrophage bactericidal activity through 819 mitochondrial ROS. Nature. 2011;472(7344):476-80. 820 32. Wen H, Ma H, Cai Q, Lin S, Lei X, He B, Wu S, Wang Z, Gao Y, Liu W, et al. Recurrent 821 ECSIT mutation encoding V140A triggers hyperinflammation and promotes hemophagocytic 822 syndrome in extranodal NK/T cell lymphoma. Nat Med. 2018. 823 33. Timmons JA, Volmar CH, Crossland H, Phillips BE, Sood S, Janczura KJ, Tormakangas T, 824 Kujala UM, Kraus WE, Atherton PJ, et al. Longevity-related molecular pathways are subject 825 to midlife "switch" in humans. Aging Cell. 2019;18(4):e12970. 826 34. Sarkar S, Leaman DW, Gupta S, Sil P, Young D, Morehead A, Mukherjee D, Ratliff N, Sun Y, 827 Rayborn M, et al. Cardiac overexpression of myotrophin triggers myocardial hypertrophy and 828 heart failure in transgenic mice. J Biol Chem. 2004;279(19):20422-34. 829 35. Harris SP, Bartley CR, Hacker TA, McDonald KS, Douglas PS, Greaser ML, Powers PA, and 830 Moss RL. Hypertrophic cardiomyopathy in cardiac myosin binding protein-C knockout mice. 831 Circ Res. 2002;90(5):594-601. 832 36. Iuso A, Repp B, Biagosch C, Terrile C, and Prokisch H. Assessing Mitochondrial Bioenergetics 833 in Isolated Mitochondria from Various Mouse Tissues Using Seahorse XF96 Analyzer. 834 Methods Mol Biol. 2017;1567(217-30. 835

836

837 Figures 838

839 840 Figure 1. Humanised ECSIT mice show intact activation of NFB and MAPK pathways 841 but impaired production of mROS and clearance of S. Typhimurium. (A) Immunoblot 842 analysis of ECSIT, -actin and total and phosphorylated (p) forms of IB, JNK, ERK and 843 p38 MAPK in cell lysates isolated from WT and ECSIT+/+ BMDMs after LPS stimulation 844 (100 ng/ml) for the indicated time periods. (B, C) ELISA analysis of (B) IL6 and (C) TNF 845 in cell supernatants of WT and ECSIT+/+ BMDM after LPS stimulation (100 ng/ml) for 0-24h. 846 (D) Immunoblot analysis of ECSIT, -actin and total and phosphorylated (p) forms of IB, 847 JNK, ERK and p38 MAPK in cell lysates isolated from WT and ECSIT+/+ BMDMs after 848 challenge with heat-inactivated S. Typhimurium (strain SL1344) for indicated times at an 849 estimated multiplicity of infection (MOI) of 10. (E, F) ELISA analysis of (E) IL-6 and (F) 850 TNF in cell supernatants of WT and ECSIT+/+ BMDM post infection. (G) MitoSox (mROS) 851 and CM-H2DCFDA (cROS) staining of WT and ECSIT+/+ BMDMs infected S. Typhimurium 852 with mean fluorescence intensity (MFI) expressed relative to uninfected WT cells. (H) CFU

853 analysis of BMDMs from WT and ECSIT+/+ mice after 2-24 h infection with S. Typhimurium 854 SL1344 at multiplicity of infection (MOI) of 10 and 50. (I) CFU analysis of homogenised liver 855 and spleen tissue from WT, Ecsit+/- and ECSIT+/+ mice post infection for 72h by oral gavage 856 with S. Typhimurium SL1344 (1 x 107 CFU). Data are expressed as means ± SD from at least 857 3 technical replicates and each representation has at least 2 independent experiments (A-H); 858 unpaired 2-tailed Student’s t test, *p<0.05, **p<0.01. Data indicate samples from individual 859 mice (n=5-8) (I); one-way ANOVA with Dunnett’s multiple comparisons test, **p<0.01. 860

861

862

863 Figure 2. Humanised ECSIT mice develop cardiac hypertrophy leading to premature 864 death. (A) Survival rates from a cohort of age- and sex-matched wild type (WT), heterozygous 865 (ECSIT+/-) and homozygous (ECSIT+/+) mice (n=20) over a 12-month time period. 866 ****p<0.0001 represents a Mantel-Cox log-rank test of survival analysis. (B) Representative 867 images of hearts from male and female WT and ECSIT+/+ mice at 7-months of age. (C) Ratios 868 of tissue weight (TW):body weight (BW) of WT and ECSIT+/+ mice (n=4-8) at 8-10 weeks. 869 (D) Echocardiograph scans of WT, ECSIT+/- and ECSIT+/+ mice (n=8) at 6-7 months of age 870 measuring Left Ventricular End Diastolic Diameter (LVEdD), Left Ventricular End systolic 871 Diameter (LVEsD), Fractional shortening (FS) and Heart rate. (E) Representative H&E, WGA 872 and Trichrome staining of left ventricular heart tissue from 6-7 months WT, ECSIT+/- and 873 ECSIT+/+ mice. (F) Quantification of WGA determined cell area and Trichrome based fibrosis 874 scores; n=10 and 18 respectively. (C, D, F) Data are expressed as means ± SD; one-way 875 ANOVA with Dunnett’s multiple comparisons test,*p<0.05, **p<0.01, ***p<0.001, 876 ****p<0.0001. Scale bar (E) = 20 μm (H&E); 50 μm (WGA); 20 μm (Trichrome).

877

878 Figure 3. ECSIT negatively correlates with human left ventricular hypertrophy and 879 cardiac fibrosis. (A) Left ventricular mass index. (B) Masson's Trichome Collagen volume 880 fraction. (C) negative correlations between ECSIT IHC lavg and markers of collagen turnover; 881 PICP+PINP/CITP ratio r= -0.38, p=0.019. (D) negative correlations between ECSIT mRNA 882 and Collagen volume fraction (CVF). (E) negative correlations between ECSIT mRNA and 883 human cardiac procollagen C-proteinase enhancer (PCPE) mRNA and procollagen C- 884 proteinase (PCP) mRNA. (F) negative correlations between ECSIT mRNA and human cardiac 885 Lysyl Oxidase (LOX) mRNA. The study utilized a cohort of 38 patients. The relationships 886 between ECSIT and markers of collagen turnover were assessed using Pearson or Spearman 887 rank order correlations for variables that were normally or non-normally distributed 888 respectively. 889

890

891

892 Figure 4. Proteomic analysis of humanised hECSIT heart. (A) Volcano plot analysis of 893 differentially expressed proteins in heart tissue from 3 and 10-month old WT and ECSIT+/+ 894 mice. (B) Heat-map analysis of differentially expressed proteins in heart tissue from 3- and 10- 895 month old WT versus ECSIT+/+ mice. (C) Heat-map analysis of differentially expressed 896 oxidative phosphorylation associated mitochondrial proteins in heart tissue from 3- and 10- 897 month old WT versus ECSIT+/+ mice. (D) STRING network analysis of statistically significant 898 differentially expressed proteins of decreased abundance in heart tissue from 3-month old and 899 10-month old WT and ECSIT+/+ mice. Mitochondrial proteins are highlighted in red. FDR 900 (false discovery rate) are computed p-values corrected for multiple testing using the method of 901 Benjamini and Hochberg. Data indicate samples from individual mice (n=3-5 for all groups). 902 (E) Immunoblot analysis of left ventricular cardiac tissue for levels of ECSIT, NDUFS3,

903 CoxIV, CoxV, Tom20 and GAPDH in WT and ECSIT+/+ mice (n=3) of 8-10 weeks age. (F, 904 G) Isolated heart mitochondria from WT and ECSIT+/+ mice at (F) 7 months (n=2) and (G) 10 905 weeks age (n=2) were subjected to BN-PAGE followed by in-gel activity assays using complex 906 I and complex IV-specific substrates. Complex I activity, as an individual complex or as part 907 of supercomplexes (SC), is shown in violet. Complex IV activity, as an individual complex or 908 as part of supercomplex with complex III is shown in brown. 909

910

911

912 Figure 5. Heart from human ECSIT knock-in mice show increased fission and 913 fragmented mitochondria. (A) Electron microscopy of the left ventricular tissue from 6- 914 month old female and 7-month old male WT and ECSIT+/+ mice (scale bars=500 nm). Red 915 arrows indicate mitochondria with distorted or absent cristae. Blue arrows indicate small and 916 fragmented mitochondria. (B) Mean long and short axes lengths (nm) of mitochondria (n=190) 917 in WT and ECSIT+/+ mice. (C) Immunoblot analysis of left ventricular cardiac tissue for levels 918 of ECSIT, phosphorylated –DRP1 (p-DRP1), DRP1, Fis1, mitofusin 1 (MFN1), mitofusin 2 919 (MFN2), OPA-1, p62, Parkin, PINK-1 and GAPDH in WT and ECSIT+/+ mice of 8-10 weeks 920 age (n=3). (D) Quantitative analysis of mitochondria dynamics and mitophagy proteins (n=3, 921 individual mouse hearts). (E) Ratio of mitochondrial DNA encoding COI:nuclear DNA 922 encoding ndufv1 as determined by quantitative PCR of cardiac tissue from 7-month old WT 923 and ECSIT+/+ mice (n = 3, individual mouse hearts). (B, D, E) Data are expressed as means ± 924 SD from 3 independent experiments; unpaired 2-tailed Student’s t test. *p<0.05, **p<0.01, 925 ***p<0.001, ****p<0.0001.

926

927 Figure 6. Knockdown of hECSIT expression in cardiomyocytes impairs mitochondrial 928 complex I activity and promotes cellular hypertrophy. (A) Immunoblot analysis, using anti- 929 ECSIT and anti-actin antibodies, of cell lysates from AC16 cardiomyocyte cells transfected 930 with control (Ctrl) or hECSIT specific siRNA (40 nM). (B, C) AC16 cardiomyocytes 931 transfected with control or hECSIT specific siRNA and then treated with PBS or Angiotensin 932 II (Ang II; 2 µM) for 24 h. (B) Cells were stained for -Actinin staining by 933 immunofluorescence microscopy with nuclei being visualized by DAPI staining. Scale bars, 934 50 µm. (C) Quantiﬁcation of cell surface area of AC16 cells (100 cells were measured for each 935 treatment from three independent experiments). (D) Quantiﬁcation of cell surface area of AC16 936 cells (100 cells were measured for each treatment from two independent experiments) 937 transfected with control or hECSIT specific siRNA, allowed recover and grown for 7 days and 938 then treated with PBS or Angiotensin II (Ang II; 2 µM) for 24 h. Immunoblot analysis, using 939 anti-ECSIT, anti-NUDFS3 and anti-GAPDHantibodies, of cell lysates. (E) Isolated heart 940 mitochondria from AC16 cells, previously transfected with control or hECSIT specific siRNA, 941 were subjected to BN-PAGE followed by in-gel activity assays using complex I and complex 942 IV-specific substrates. Complex I activity, as part of supercomplexes (SC), is shown in violet. 943 Complex IV activity, as an individual complex or as part of supercomplex with complex III is 944 shown in brown. (F) Immunoblot analysis of lysates from AC16 cells, previously transfected 945 with control or hECSIT specific siRNA, for levels of ECSIT, p62, PINK1, NDUFS3, Cox IV, 946 Tom 20, Fis1, mitofusin 1 (MFN1), mitofusin 2 (MFN2), OPA-1 and GAPDH. Data represent

947 at least two biological replicates. (C, D) Data are expressed as means ± SD; one-way ANOVA 948 with Dunnett’s multiple comparisons test; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. 949 950

951

952

953 Figure 7. Reduced levels of hECSIT causes a metabolic shift from oxidative 954 phosphorylation to glycolysis (A) Analysis of oxygen consumption rate (OCR) in AC16 cells 955 transfected with control and hECSIT siRNA to assess rates of basal respiration, maximal 956 respiration, respiration linked to ATP production, proton leak and spare respiratory capacity. 957 (B) Analysis of extracellular acidification rate (ECAR) in AC16 cells transfected with control 958 and hECSIT siRNA to assess non-glycolytic acidification, glycolysis, glycolytic capacity and 959 glycolytic reserve. (C) Glycolysis driven proton efflux rate in AC16 cells transfected with 960 control and hECSIT siRNA. (D) Rates of ATP production as mediated by glycolysis or 961 mitochondrial metabolism in AC16 cells transfected with control and hECSIT siRNA (left 962 panel). Ratio of ATP produced by oxidative phosphorylation:ATP produced by glycolysis 963 ATP in AC16 cells transfected with control and hECSIT siRNA (right panel). Data are 964 expressed as means ± SD from 4-6 technical replicates, and each experiment has at least two 965 biological replicates; unpaired 2-tailed Student’s t test applied. *p<0.05, **p<0.01, 966 ***p<0.001, ****p<0.0001. 967

968 969 Figure 8. Reprogramming of primary metabolism to glycolysis in humanised hECSIT 970 heart. (A) Analysis of oxygen consumption rate (OCR) in mitochondria isolated from WT and 971 ECSIT+/+. (B) Quantitative data of basal respiration, maximal respiration, ATP production and 972 proton leak in mitochondria isolated from WT and ECSIT+/+. (C) Levels of ATP in left 973 ventricular tissue from hearts of WT and ECSIT+/+ mice of 6-7 months old. Data are expressed 974 as means ± SD (n=6, two pair mice, three technical repeats for each pair, gender mixed); * 975 p<0.05, paired t test. (D) Levels of lactate in left ventricular tissue from hearts of WT and +/+ 976 ECSIT mice (n=4) of 7-months old. Data are expressed as means ± SD . (E) Levels of H2O2 977 in left ventricular tissue from hearts of WT and ECSIT+/+ mice of 6-7 months old. Data are 978 expressed as means ± SD (n=6, two pair mice, three technical repeats for each pair, gender 979 mixed); * p<0.05, unpaired t test. (A-B) Data are expressed as means ± SD from 4-6 technical 980 replicates, and each experiment has two biological replicates; unpaired 2-tailed Student’s t test 981 applied. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. 982 983 984 985 986